INDEX

Explanations

examples and codes from technical documentation

oai_token-act-pair · gpt-3.5-turbo

references to different examples within a context

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 millenn

-0.83

kefeller

-0.74

lock

-0.74

ternity

-0.73

ippery

-0.73

orne

-0.73

emetery

-0.72

resent

-0.71

cious

-0.71

 adolescence

-0.68

POSITIVE LOGITS

 Example

0.91

 snipp

0.83

 Flask

0.83

Example

0.81

example

0.79

 dummy

0.76

 Takeru

0.76

 example

0.75

 examples

0.74

 Examples

0.73

Activations Density 0.037%

examples and codes from technical documentation

references to different examples within a context

No Comments

No Known Activations